Hybrid airborne imaging systems integrating oblique photogrammetric cameras and LiDAR sensors provide complementary geometric and radiometric information for high-fidelity 3D reconstruction. While LiDAR offers stable global geometric accuracy, photogrammetric reconstructions provide dense surface detail and rich fac¸ade information through oblique imaging. However, photogrammetric models often exhibit locally varying geometric drift caused by error accumulation during bundle adjustment, preventing pixel-accurate alignment between images and LiDAR data. Conventional global registration approaches are insufficient to correct such spatially varying misalignments. In this paper, we propose a view-dependent fusion framework for pixel-accurate registration between photogrammetric images and LiDAR data. The global alignment problem is decomposed into localized rigid registrations performed independently for each image. For each view, a local photogrammetric point cloud reconstructed from depth maps is rigidly aligned to a corresponding LiDAR subset using fixed-scale point-to-plane ICP. The estimated transformation is used to project LiDAR points into the image domain, where depth-consistency checks enforce visibility constraints and remove inconsistent measurements. Valid LiDAR points are colorized and fused with the photogrammetric reconstruction to generate anchored view-dependent point clouds. Finally, all view-dependent reconstructions are aggregated using voxel-grid fusion with medianbased point selection. Experiments on airborne oblique imaging datasets demonstrate improved reconstruction completeness and pixel-level consistency between images and LiDAR geometry. The proposed framework provides a practical and robust solution for high-accuracy multi-modal 3D reconstruction in complex urban environments.